304 
MR. W. CROOKES ON REPULSION RESULTING PROM RADIATION. 
and the vanes rotated rapidly in the positive direction, as if driven round by 
molecular pressure coming from the hot wire. The rotation kept up with great 
rapidity as long as the wire was kept hot, and did not show any signs of diminution 
of speed. 
When the apparatus was again cold, and the vanes quiet, I put a finger on the top 
plate of glass so as to cause molecular pressure to strike the vanes from above down- 
wards. The vanes now moved in the negative direction, / or normally to the 
source of pressure. 
The lower part of the apparatus was now grasped in the hand to warm it, when 
positive rotation commenced, showing that pressure came from beneath. 
339. Air was now admitted into the apparatus until the gauge was depressed 
12 millims. Battery contact was made, and very slow negative rotation immediately 
took place, / at the rate of one revolution in 168 seconds. The exhaustion was 
continued, followed by a gradual increase in the speed up to about 400 millionths, 
at which point the rate was 10 revolutions a minute, still in the negative direction. 
A little beyond this degree of exhaustion, the vanes refused to move when the platinum 
wire was heated. At a higher rarefaction, positive rotation, took place. At a 
rarefaction of 34 millionths of an atmosphere (about the point of maximum sensitive- 
ness for a radiometer), the speed of the vanes was 200 revolutions a minute. At 
3 millionths of an atmosphere the speed was 300 revolutions a minute ; and at 
1 millionth of an atmosphere the speed continued about the same. Owing to the 
cement joint I was unable to get a greater amount of rarefaction with this apparatus. 
340. These results are of interest in many respects. When using a candle as the 
source of radiation, I have always found very little repulsion until the gauge has risen 
to within about 5 millims. of a vacuum ; from this point the repulsion increases 
steadily up to a rarefaction of about 35 millionths of an atmosphere, when it rapidly 
sinks, until at O'l millionth it is less than one-tenth of its maximum.* Below 
5 millims. attraction or repulsion takes place according to circumstances which are not 
clearly explained ; but above that point, when repulsion has fairly commenced, I have 
never observed a change of sign. In the experiments just described, the radiation from 
an ignited platinum wire is used ; there are only about 5 millims. space between the wire 
and the fly, and no glass intervenes. The results, therefore, are likely to be much 
more definite than with the usual radiometer-kind of apparatus, and the actions should 
commence at a lower pressure. 
341. In previous experiments with candles, the abnormal movement at low exhaus- 
tions was faint and irregular ; here they occur with a sharpness which gives one hope 
of getting at a law. The negative rotation of the fly is evidently the analogue of the 
attraction observed in my early experiments at low exhaustions, both being abnormal 
* Proc. Roy. Soc., Nov. 16, 1876, No. 175, p. 305. 
